ATProto Browser
Experimental browser for the Atmosphere
Experimental browser for the Atmosphere
Main result #3. Insufficiency of training-time interventions. We show that the use of inference-time compute above is fundamental, in the sense that purely training time interventions (e.g., modifications to DPO) that produce proper policies cannot achieve similar guarantees in polynomial time. 15/
Mar 27, 2025, 5:28 PM
{
"uri": "at://did:plc:x2a3inabvfsn4wntrlbbndrv/app.bsky.feed.post/3llet5ubujf2c",
"cid": "bafyreiahice3ju7iv3q2uqg3ys6w6eddb3dfuiesesicpqgssqldnbq2iu",
"value": {
"text": "Main result #3. Insufficiency of training-time interventions. We show that the use of inference-time compute above is fundamental, in the sense that purely training time interventions (e.g., modifications to DPO) that produce proper policies cannot achieve similar guarantees in polynomial time.\n\n15/",
"$type": "app.bsky.feed.post",
"embed": {
"$type": "app.bsky.embed.images",
"images": [
{
"alt": "",
"image": {
"$type": "blob",
"ref": {
"$link": "bafkreihanps7vvryfe36yilzrouplo3nvtcrhbakjwtc4uecmckt5qk2ye"
},
"mimeType": "image/jpeg",
"size": 182016
},
"aspectRatio": {
"width": 1160,
"height": 250
}
}
]
},
"langs": [
"en"
],
"reply": {
"root": {
"cid": "bafyreidzv7o2gllewwfqqi4ixremjopxmuyntc5gstydqektrxkcjni4pi",
"uri": "at://did:plc:x2a3inabvfsn4wntrlbbndrv/app.bsky.feed.post/3llet5p66ac2c"
},
"parent": {
"cid": "bafyreigqmwhzfk5nz62ncjh7rpgjdwxgmtcwld6ozms624nbmsdzuwx5ym",
"uri": "at://did:plc:x2a3inabvfsn4wntrlbbndrv/app.bsky.feed.post/3llet5ubskv2c"
}
},
"createdAt": "2025-03-27T17:28:13.784Z"
}
}